WO2020253664A1 - Procédé et système de transmission vidéo, et support de stockage - Google Patents

Procédé et système de transmission vidéo, et support de stockage Download PDF

Info

Publication number
WO2020253664A1
WO2020253664A1 PCT/CN2020/096239 CN2020096239W WO2020253664A1 WO 2020253664 A1 WO2020253664 A1 WO 2020253664A1 CN 2020096239 W CN2020096239 W CN 2020096239W WO 2020253664 A1 WO2020253664 A1 WO 2020253664A1
Authority
WO
WIPO (PCT)
Prior art keywords
video data
qoe
requested
edge server
video
Prior art date
Application number
PCT/CN2020/096239
Other languages
English (en)
Chinese (zh)
Inventor
李清
石婉欣
王潮
林栋�
邓轻松
Original Assignee
鹏城实验室
南方科技大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 鹏城实验室, 南方科技大学 filed Critical 鹏城实验室
Publication of WO2020253664A1 publication Critical patent/WO2020253664A1/fr

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/23—Processing of content or additional data; Elementary server operations; Server middleware
    • H04N21/231—Content storage operation, e.g. caching movies for short term storage, replicating data over plural servers, prioritizing data for deletion
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/23—Processing of content or additional data; Elementary server operations; Server middleware
    • H04N21/239—Interfacing the upstream path of the transmission network, e.g. prioritizing client content requests
    • H04N21/2393—Interfacing the upstream path of the transmission network, e.g. prioritizing client content requests involving handling client requests
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/25—Management operations performed by the server for facilitating the content distribution or administrating data related to end-users or client devices, e.g. end-user or client device authentication, learning user preferences for recommending movies
    • H04N21/262—Content or additional data distribution scheduling, e.g. sending additional data at off-peak times, updating software modules, calculating the carousel transmission frequency, delaying a video stream transmission, generating play-lists
    • H04N21/26208—Content or additional data distribution scheduling, e.g. sending additional data at off-peak times, updating software modules, calculating the carousel transmission frequency, delaying a video stream transmission, generating play-lists the scheduling operation being performed under constraints
    • H04N21/26216—Content or additional data distribution scheduling, e.g. sending additional data at off-peak times, updating software modules, calculating the carousel transmission frequency, delaying a video stream transmission, generating play-lists the scheduling operation being performed under constraints involving the channel capacity, e.g. network bandwidth
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/60—Network structure or processes for video distribution between server and client or between remote clients; Control signalling between clients, server and network components; Transmission of management data between server and client, e.g. sending from server to client commands for recording incoming content stream; Communication details between server and client 
    • H04N21/63—Control signaling related to video distribution between client, server and network components; Network processes for video distribution between server and clients or between remote clients, e.g. transmitting basic layer and enhancement layers over different transmission paths, setting up a peer-to-peer communication via Internet between remote STB's; Communication protocols; Addressing
    • H04N21/643—Communication protocols
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/60—Network structure or processes for video distribution between server and client or between remote clients; Control signalling between clients, server and network components; Transmission of management data between server and client, e.g. sending from server to client commands for recording incoming content stream; Communication details between server and client 
    • H04N21/63—Control signaling related to video distribution between client, server and network components; Network processes for video distribution between server and clients or between remote clients, e.g. transmitting basic layer and enhancement layers over different transmission paths, setting up a peer-to-peer communication via Internet between remote STB's; Communication protocols; Addressing
    • H04N21/647—Control signaling between network components and server or clients; Network processes for video distribution between server and clients, e.g. controlling the quality of the video stream, by dropping packets, protecting content from unauthorised alteration within the network, monitoring of network load, bridging between two different networks, e.g. between IP and wireless
    • H04N21/64723—Monitoring of network processes or resources, e.g. monitoring of network load
    • H04N21/64738—Monitoring network characteristics, e.g. bandwidth, congestion level

Definitions

  • This application relates to the field of computer network technology, in particular to a video transmission method, system and storage medium.
  • Dynamic adaptive streaming technology (Dynamic Adaptive Streaming over HTTP, DASH) based on HyperText Transfer Protocol (HTTP) is widely used due to its flexibility and compressibility. This transmission method can improve video transmission. Bandwidth utilization, thereby improving user experience.
  • DASH Dynamic Adaptive Streaming over HTTP
  • HTTP HyperText Transfer Protocol
  • the client requests content based on the video block sequence and adopts a bit rate adaptive algorithm locally, adaptively selects the most appropriate bit rate according to the real-time network status, and automatically adjusts the download and playback plan according to the current network conditions. Avoid playing jams or rebuffering events, thereby improving the overall transmission efficiency of the network, seamlessly adapting to the constantly changing network and providing a high-quality playback experience, making the network video playback smoother.
  • a video transmission method including:
  • the edge server receives the request signal sent by the client, and extracts characteristic information of the request signal; the request signal is used to instruct to obtain the video data to be requested;
  • the edge server searches the local database for the video data to be requested according to the characteristic information of the request signal; the local database stores prefetched videos and cached videos; and
  • the edge server transmits the video data to be requested to the client.
  • the method further includes:
  • the edge server predicts the QOE gain value of the current video experience quality according to the characteristic information of the request signal.
  • the edge server performs prefetch management and cache management of video data according to the current video QOE gain value and the characteristic information of the request signal.
  • the edge server includes a QOE gain predictor
  • the edge server predicts the current video QOE gain value according to the characteristic information of the request signal, including:
  • the QOE gain predictor predicts the cache hit throughput and the cache miss throughput of the video data to be requested according to the characteristic information of the request signal
  • the QOE gain predictor predicts the current video QOE gain value according to the cache hit throughput and the cache miss throughput.
  • the QOE gain predictor predicting the current QOE gain value according to the cache hit throughput and the cache miss throughput includes:
  • the QOE gain predictor adopts a preset neural network model, and respectively presets the QOE value corresponding to the cache hit throughput and the QOE value corresponding to the cache miss throughput;
  • the QOE gain predictor determines the difference between the QOE value corresponding to the cache hit throughput and the QOE value corresponding to the cache miss throughput as the current QOE gain value.
  • the edge server prefetching and caching video data according to the current video QOE gain value and the characteristic information of the request signal includes:
  • the edge server performs buffer management on the to-be-requested video data according to the current video QOE gain value, and performs a buffer management on the to-be-requested video data according to the current video QOE gain value and the characteristic information of the request signal. Associate video data for prefetch management.
  • the edge server includes a cache manager
  • the edge server performs cache management on the video data to be requested according to the current video QOE gain value, including:
  • the buffer manager obtains the accumulated QOE gains of all the video data to be requested in the first half of the buffer period;
  • the cache manager sorts the cumulative QOE gain values of each video data to be requested from high to bottom, and caches the video data to be requested corresponding to the cumulative QOE gain values in the order from high to bottom until the cache manager Full.
  • the edge server includes a prefetch manager
  • the edge server performs prefetch management on the associated video data of the video data to be requested according to the current video QOE gain value and the characteristic information of the request signal, including:
  • the prefetch manager obtains multiple video data associated with the video data to be requested according to the characteristic information of the request signal
  • the prefetch manager predicts the maximum benefit value of the associated video data that can be prefetched in the current prefetch period
  • the prefetch manager performs prefetch management on the associated video data according to the maximum benefit value of the associated video data, the current video QOE gain value, and each of the associated video data.
  • the prefetch manager performs prefetch management on the associated video data according to the maximum benefit value of the associated video data, the current video QOE gain value, and each of the associated video data, include:
  • the prefetch manager determines the video data to be prefetched according to the maximum benefit value of the associated video data and the current video QOE gain value;
  • the prefetch manager sends a prefetch request carrying the video data to be prefetched to the source server;
  • the prefetch manager receives the video data to be prefetched returned by the source server, and stores the video data to be prefetched.
  • a video transmission system which is applied to the above-mentioned video transmission method, and the system includes: a client, an edge server, and an origin server;
  • the client terminal is configured to send and obtain the video data to be requested to the source server and receive the video data to be requested; wherein the data to be requested is returned by the edge server according to the request signal redirected by the source server;
  • the origin server is configured to redirect the request signal to the edge server; or, when the to-be-requested video data is not stored in the edge server, send the to-be-requested video data to the client;
  • the edge server is used to execute:
  • the edge server receives the request signal sent by the client, and extracts characteristic information of the request signal; the request signal is used to instruct to obtain the video data to be requested;
  • the edge server searches the local database for the video data to be requested according to the characteristic information of the request signal; the local database stores prefetched videos and cached videos; and
  • the edge server transmits the video data to be requested to the client.
  • a computer-readable storage medium on which a computer program is stored, and the computer program is executed when the computer program is executed by a processor:
  • the edge server receives the request signal sent by the client, and extracts characteristic information of the request signal; the request signal is used to instruct to obtain the video data to be requested;
  • the edge server searches the local database for the video data to be requested according to the characteristic information of the request signal; the local database stores prefetched videos and cached videos; and
  • the edge server transmits the video data to be requested to the client.
  • Fig. 1 is a schematic diagram of a video transmission system provided by an embodiment.
  • Fig. 2 is a schematic structural diagram of an edge server provided by an embodiment.
  • Fig. 3 is a schematic flowchart of a video transmission method provided by an embodiment.
  • Fig. 4 is a schematic flowchart of a video transmission method provided by an embodiment.
  • Fig. 5 is a schematic flowchart of a video transmission method provided by an embodiment.
  • Figure 5a is a scatter diagram of the relationship between throughput and other indicators provided by an embodiment.
  • Fig. 6 is a schematic flowchart of a video transmission method according to an embodiment.
  • Fig. 6a is a flow chart of throughput and QoE prediction provided by an embodiment.
  • Fig. 6b is a schematic diagram of video classification of SSIM provided by an embodiment.
  • Fig. 7 is a schematic flowchart of a video transmission method provided by an embodiment.
  • Fig. 8 is a schematic flowchart of a video transmission method provided by an embodiment.
  • Fig. 8a is a schematic diagram of chain length change probability according to an embodiment.
  • Fig. 9 is a schematic flowchart of a video transmission method provided by an embodiment.
  • FIG. 10 is a schematic flowchart of a video transmission method provided by an embodiment.
  • Figure 11 is an internal structure diagram of a computer device in an embodiment.
  • DASH technology is adopted in the prior art, and video transmission still has technical problems of low performance and efficiency.
  • the video transmission method provided by this application can be applied to the video transmission system as shown in FIG. 1.
  • the system includes a client, an edge server, and an origin server; in this system, all video resources are stored on the origin server, The hot content is stored in the edge server.
  • the origin server and the edge server cooperate with each other. After receiving the client request, the origin server will redirect the request to the edge server. If the content requested by the client is In the edge server, the edge server will directly return the content to the client. If the content requested by the client is not in the edge server, the edge server receives the video sent by the source server through the wide area network and sends the video to the client.
  • the WAN can be simplified and modeled as a bottleneck link from the source server to the edge server.
  • the client can be a smart device that can access the Internet, such as a smart phone, a computer, or an IPAD, which is not limited in this embodiment.
  • the edge server is deployed at a location closer to the client than the source server in actual applications. The specific deployment method and the implementation of this solution will be described in the following examples.
  • the dynamic adaptive streaming over HTTP (DASH) technology based on the HyperText Transfer Protocol (HTTP) is due to its flexibility and compressibility.
  • this transmission method can improve the bandwidth utilization of video transmission, thereby improving the quality of experience (QoE) of users.
  • the client requests content based on the video block sequence and adopts a bit rate adaptive algorithm locally, and adaptively selects the most suitable bit rate according to the real-time state of the network to meet user needs. Therefore, strengthening QoE-oriented video transmission is of great significance for improving the overall efficiency of the network.
  • DASH technology since the current network is highly dynamic, only non-guaranteed services can be provided.
  • video transmission is highly centralized in both time and space dimensions, that is, users of certain specific networks will often request some popular videos within a certain period of time. For example, Facebook’s top 0.1% and top 1% videos account for 62% and 83% of its video transmissions, respectively. The content frequently requested by users will be transmitted multiple times, causing bandwidth consumption, which leads to a large amount of waste of already sufficient bandwidth. As a result, video transmission still faces severe challenges, including problems such as unstable user experience caused by network status changes, and low bandwidth utilization efficiency caused by redundant video transmission.
  • the first type of method is the client's adaptive code rate adjustment and prediction mechanism.
  • each client makes its own code rate decision to compete for shared bandwidth.
  • This type of solution lacks overallity and fairness.
  • Some methods such as MPC (Model Predictive Control, which represents a client-side adaptive algorithm for video transmission, using control theory methods), assuming accurate prediction of network throughput, and using control theory methods to select the appropriate bit rate on the client side To get the best QoE.
  • Pensieve uses a reinforcement learning model to determine the bit rate.
  • the Bola scheme utilizes Lyapunov technology to achieve partial optimization of bit rate selection.
  • Oboe is a C++ library that is used to build high-performance audio applications with the lowest possible delay on 99% of Android devices. It is a solution that automatically adjusts the video transmission adaptive algorithm and can adapt to various states) , It will automatically adjust parameters according to different network conditions and dynamically adjust parameters according to the current network.
  • These client solutions help end users adapt to DASH video transmission, while considering buffer occupancy and transmission rate.
  • each client will make its own bit rate decision and compete for shared bandwidth.
  • the second type of method is server-side global optimization. This solution is very difficult for the server to perceive the entire network status in real time. Therefore, supporting online global optimization is a challenge because it is far away from the end user.
  • the server-based method releases the traffic load of the source server and improve the response speed to a certain extent, they require high investment in equipment and maintenance due to the large amount of computing and storage. Therefore, the server-based method is not ideal for its coarse-grained optimization effect and high cost.
  • the third type of method is based on the network-side caching and prefetching scheme.
  • Some offline caching schemes have designed complex algorithms to cache the most popular Content.
  • an online caching scheme with advanced auxiliary means such as popular content replacement algorithm based on Markov model, caching method based on enhanced learning method, and real-time dynamic caching of non-popular content is proposed.
  • the embodiments of the present application provide a video transmission method, device, computer equipment, and storage medium, aiming to solve the technical problem of low performance and efficiency in video transmission using DASH technology in the prior art.
  • the technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems will be described in detail through the embodiments and the accompanying drawings.
  • the following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.
  • the execution body of FIGS. 3-9 is an edge server
  • the execution body of FIG. 10 is a client.
  • the execution body of FIGS. 3-10 may also be A video transmission device, where the device can be implemented as part or all of video transmission through software, hardware, or a combination of software and hardware.
  • An embodiment of the present application provides a video transmission system, which includes: a client, an edge server, and an origin server; optionally, the edge server includes: a request collector, a QoE gain predictor, and a cache manager And prefetch manager.
  • the client is used to send a request signal carrying the video data to be requested to the origin server;
  • the origin server is used to redirect the request signal to the edge server; or, in the video data to be requested
  • the edge server is configured to send the to-be-requested video data to the client according to the request signal.
  • the edge server is an intelligent edge architecture in practical applications, which is used to process DASH-based video requests. It can make QoE-friendly cache prefetching decisions based on user information and network status.
  • Figure 2 presents the composition of the edge server, which consists of four modules, including request collector, cache manager, QoE gain predictor and prefetch manager.
  • the request collector is used to collect and analyze the requests from users, and send user information to the QoE gain predictor and prefetch manager respectively;
  • the QoE gain predictor uses user information and network status to predict throughput and Current video QoE gain value;
  • Cache manager used for the current video QOE gain value, to realize the functions of cache update, cache lookup and content response;
  • Prefetch manager used to calculate the benefits of prefetched video blocks and based on the calculation results To implement the prefetch strategy.
  • These modules cooperate and interact with each other to provide services for users covered by the intelligent edge server.
  • Such an intelligent edge server architecture can provide globally optimized transmission services, rather than being limited to local single-user optimization, so it is more conducive to efficient use of limited network bandwidth.
  • differentiated services can be realized intelligent differentiated services for different users or videos (even at the video block level).
  • differentiated services will not bring the burden of transmission redundancy, but can help the intelligent edge server to combat network jitter in real time, and ultimately ensure that users get a smooth viewing experience.
  • the embodiment of the present application provides an example, for example, the edge service solution is realized by the Apache Traffic Server (ATS).
  • ATS is a high-performance open source HTTP proxy, which contains a series of important state nodes in the process of processing HTTP transactions.
  • the ATS plug-in connects to the corresponding state through various HOOKs. When the HTTP transaction reaches a certain state, the corresponding HOOK will automatically call the corresponding plug-in to combine the user-defined function with the basic HTTP transaction process.
  • the four modules included in this solution can be realized through plug-ins in ATS: request collector, cache manager, prefetch manager, and QoE gain predictor.
  • CACHE_LOOKUP_COMPLETE_HOOK implements a cache manager to realize cache status feedback and QoE-based cache replacement.
  • the QoE gain predictor gets input from the function caller.
  • CNN convolutional neural network
  • the function of neural network is realized by libtorch, a C++ interface of PyTorch.
  • the QoE gain predictor is a compiled C++ program called by a system command before the request transaction of the prefetch manager module starts.
  • the historical throughput data in HSDPA is used as the bandwidth change to train the prediction model, and the QoE values of cache hits and cache misses are obtained respectively.
  • the prefetch manager processes the generated prefetch requests and sends a part of these requests to the origin server.
  • the download throughput of the source server can be continuously updated by deploying TXN_CLOSE_HOOK.
  • the edge server should be deployed closer to the user.
  • campus and corporate networks can deploy this service at an exit point connected to a network provider. This can reduce the flow of campus or enterprise users, thereby improving the quality of service.
  • video providers can also deploy the service in the form of containers, virtual machines, or applications on public cloud platforms, ISP's virtual edge data centers (DC), or some super clients. In this way, content providers can also improve the QoE of their users and reduce the bandwidth required for video transmission. Therefore, the edge server can be deployed in a variety of ways, such as deploying an x86LEAP server at the campus network exit, running the intelligent edge application on a specific client, or implementing the virtualized network function in an ISP edge data center.
  • FIG. 3 provides a video transmission method. This embodiment relates to the specific process of the edge server sending the video data to be requested according to the client's request signal. As shown in FIG. 3, the method includes:
  • the edge server receives a request signal sent by a client, and extracts characteristic information of the request signal; the request signal is used to instruct to obtain video data to be requested.
  • the request signal is a DASH request sent by the client to the source server to obtain the video data to be requested.
  • the source server will send the request signal after receiving the request signal sent by the client.
  • the edge server extracts characteristic information from the request signal.
  • the feature information can indicate user information and network status.
  • the feature information includes but is not limited to the QoE of the previous video block, the buffer progress of the player, the currently requested video block bit rate, video type, and round trip delay ( Round-Trip Time (RTT), video block chain length, throughput change, download time, video block size, and actual hits.
  • RTT Round-Trip Time
  • the QoE of the previous video block, the buffer progress of the player, the currently requested video block rate, the video type, and the RTT feature are all one-dimensional features.
  • the video type refers to the video to which the currently played video belongs based on the SSIM classification.
  • Type, and RTT refers to the round-trip time between the user and the edge server.
  • throughput changes, download time, video block size, and actual hits are all multidimensional variables.
  • the download throughput, download time, video block size, and hits of the past five video blocks can be obtained for neural network training, for example .
  • These characteristic information are as follows: lastQoE: the QoE of the last video block sensed by the client; buffer: the local cache utilization of the client, which can indicate the urgency of the client request; bitrate: the bitrate of the video block currently requested by the client Size; videoType: Classification of videos based on SSIM features.
  • chainLength the length of the video block chain formed by a user requesting a certain video, which represents the number of video blocks without bit rate conversion
  • throughputList_n downloading the past Historical throughput of n video blocks
  • downloadTime_n historical download time when downloading the past n video blocks
  • segmentSize_n the size of the n video blocks requested in the past
  • ifHit_n the hit situation of the past n requests on the edge server.
  • the present invention returns the hit information to the user, and the client maintains the hit state information
  • RTT the round-trip time between the client and the edge server. Since the available bandwidth from different users to the edge server is different and will change over time, it is necessary to feed back the link state perceived by the client to the edge server. This also avoids the maintenance overhead of the edge server on the transmission state.
  • the dimension of each feature information can be adjusted appropriately based on the calculation amount and prediction accuracy, which is not limited in this embodiment.
  • the manner in which the edge server parses the above-mentioned characteristic information from the request signal may be parsed through a protocol, or may be other manners, which is not limited in this embodiment.
  • the edge server can extract the following status information from the DASH request through the request collector for further use. These additional information only consumes a small amount of transmission bandwidth, so the additional overhead in the DASH scenario can be ignored.
  • the edge server searches the local database for the video data to be requested according to the characteristic information of the request signal; the local database stores prefetched videos and cached videos.
  • the edge server searches the local database for the video data to be requested, where the local database represents the edge server's database of prefetched videos and cached videos, where the prefetched video Indicates a video that has never been requested by the client.
  • the prefetched video is determined by the edge server in advance based on the client's usual request data analysis, and then obtained from the edge server and stored.
  • the cached video represents the video that the client has requested, and the edge server will store it. It is understandable that both the pre-fetched video and the cached video are hot content with high popularity, that is, it can basically realize that all the videos that users usually need to watch are stored in the local database of the edge server.
  • the edge server transmits the video data to be requested to the client.
  • the edge server after finding the video requested by the client in the local database, the edge server directly transmits the video data to be requested to the client. It should be noted that if the video data to be requested does not exist in the local database of the edge server, the edge server will request the video from the source server in real time, and then send the video to the client after receiving the video to be requested.
  • the video transmission method provided in this embodiment uses the edge server to pre-fetch and cache the more popular videos, so that when the client requests the video, the edge server can efficiently output the corresponding video to the client. Improved video transmission performance and efficiency.
  • the embodiment of the present application also provides a video transmission method, which involves the specific process of storing prefetched video and buffering video on the edge server.
  • the above S102 step includes:
  • the edge server predicts the QOE gain value of the current video experience quality according to the characteristic information of the request signal.
  • the edge server needs to predict the current video based on the characteristic information of the request signal. Quality of experience QOE gain value.
  • the method for the edge server to predict the QOE gain value of the current video experience quality may be to use a deep neural network for prediction.
  • all the feature information of the request signal extracted above is input into a pre-trained neural network model, and then, According to the output of the neural network model, the current video experience quality QOE gain value can be directly determined.
  • other methods can also be used. For example, first determine the throughput when the cache hits and the throughput when the cache misses, and then according to the preset The algorithm combines the throughput when the cache hits and the throughput when the cache misses to determine the QOE gain value of the current video, which is not limited in this embodiment, and can be determined according to actual conditions.
  • the edge server performs prefetch management and cache management of the video data according to the current video QOE gain value and the characteristic information of the request signal.
  • the edge server Based on the current video QOE gain value determined in step S202, the edge server performs video data prefetch management and cache management according to the current video QOE gain value and the characteristic information of the request signal. For example, the edge server compares the QOE gain value with High video caching or prefetching. In this way, when the bandwidth of the server is redundant, the edge server prefetches the video block with the highest QoE gain. When the bandwidth is insufficient, the previously prefetched content can be used to ensure a better viewing experience for users. Similarly, the video popularity and QoE gain can be adjusted Cache, when the client requests a video, the corresponding video can be efficiently transmitted to the client, ensuring that the user gets a better viewing experience.
  • the edge server includes a QOE gain predictor; as shown in Fig. 5, the step S201 includes:
  • the QOE gain predictor predicts the cache hit throughput and the cache miss throughput of the video data to be requested according to the characteristic information of the request signal.
  • network throughput is an important factor in QoE gain prediction, so we first use linear regression to predict future throughput.
  • the hash distribution diagram shown in Figure 5a shows the correlation between throughput and other factors.
  • Each sub-graph in the hash distribution graph shows a certain trend, and different hash states indicate different correlations. If there is a regular trend between the two variables, it has a strong correlation, otherwise the correlation is weak.
  • the results show that in the case of cache hits and cache misses, the size of the video block, the RTT from the user to the edge server, and the download time of the video block are all closely related to the network throughput. In a video request, cache hits and cache misses cannot occur at the same time.
  • the QOE gain predictor can predict the throughput of cache hits and misses through linear regression.
  • the QOE gain predictor predicts the current video QOE gain value according to the cache hit throughput and the cache miss throughput.
  • an implementation manner of the foregoing S302 includes:
  • the QOE gain predictor uses a preset neural network model to respectively predict the QOE value corresponding to the cache hit throughput and the QOE value corresponding to the cache miss throughput.
  • the input of the neural network in the QOE gain predictor includes feature information mainly obtained from the request collector, including historical throughput, download time of several video blocks, historical hit status, and the relationship between the client and the server.
  • Inter-RTT will first be used in linear regression for the throughput of past several video blocks hit or not.
  • the throughput information of all video blocks in the hit state obtained by linear regression will be used with other features obtained by the request collector to predict the QOE value of the next video block hit.
  • by predicting the throughput of cache misses it can also be obtained The QOE value of the next video block miss.
  • the QOE gain predictor determines the difference between the QOE value corresponding to the cache hit throughput and the QOE value corresponding to the cache miss throughput as the current QOE gain value.
  • the QOE gain predictor determines the difference between the two as the current QOE gain value. For example, if The QOE value corresponding to the cache hit throughput is The QOE value corresponding to the cache miss throughput is expressed as Then the QoE gain of the next video block hit can be obtained and the benefit value can be calculated.
  • the formula is:
  • QoE the factors affecting QoE in this application include the average bit rate, buffering time and bit rate switching correlation. So QoE can be defined as follows:
  • q represents the quality level of the video
  • the two coefficients u and ⁇ are used to adjust the impact of buffering time and bit rate switching on the user experience.
  • structural similarity can be used to characterize the quality of different video blocks.
  • different videos can be classified into different categories according to their SSIM characteristics. The steeper the straight line in Fig. 6b indicates that the higher the QoE gain brought by the video block with the higher bit rate. In the same way, a relatively flat curve indicates that the benefits of caching video blocks with different bit rates are not much different.
  • the QoE gain prediction model is not trained on the user side, but uses the Backend server to assist in training the prediction model.
  • Edge servers are a good choice. Therefore, training QoE prediction models, calculating QoE gains, and collecting historical statistical data on edge servers can make full use of computing power and mobilize the available resources of edge servers. That is, the limited computing power and storage capacity of intelligent edges can be fully utilized and greatly improved The user experience under the dynamic network.
  • the manner in which the edge server performs video data prefetching and caching according to the current video QOE gain value and the characteristic information of the request signal in the above S202 includes the edge server performing cache management on the requested video data according to the current video QOE gain value, and According to the current video QOE gain value and the characteristic information of the request signal, there are two cases of prefetching and managing the associated video data of the video data to be requested. The following two cases are respectively described through several embodiments.
  • the process of performing cache management for the edge server where the above-mentioned edge server includes a cache manager; then S202 includes:
  • the cache manager obtains the accumulated QOE gains of all the video data to be requested in the first half of the cache period.
  • the cache manager executes the corresponding cache strategy and regularly updates the locally cached content.
  • the cache update period is defined as Tc, and it is assumed that in the sth period, all requests for video f include Hf(sTc) video block IDs, and each video block corresponds to a request quantity.
  • the buffer manager calculates the average QoE gain of all requested videos according to the video QoE gain value sent by the received QoE gain predictor, and then counts the cumulative value of the QoE gain brought by the previous video block in a period, that is, a statistics The cumulative value of the video block gain in the first half of the cycle.
  • the cache manager sorts the cumulative QOE gain values of each video data to be requested from high to bottom, and caches the to-be-requested video data corresponding to the cumulative QOE gain values in the order from high to bottom until the cache manager is full .
  • the cache manager Based on the cumulative value of the first half-period video block gain in each cycle counted by the cache manager in step S501, the cache manager sorts the cumulative QOE gain of each video data to be requested from high to bottom, and sorts QOE in the order from high to bottom.
  • the to-be-requested video data corresponding to the accumulated gain value is cached until the buffer manager is full. It is understandable that it is not necessary to sort all of them. Just compare the accumulated value of the latest QOE gain with the cached video block. Video blocks with larger gains are replaced until the buffer space is filled.
  • the cache manager caches and updates the video block with larger gain, that is, the hot content, so that all the videos that the client needs to request can be covered greatly, and the videos that satisfy the client request are all Stored in the cache manager of the edge server.
  • the process of prefetching management is performed for the edge server, where the above-mentioned edge server includes a prefetching manager; then S202 includes:
  • the prefetch manager obtains multiple video data associated with the video data to be requested according to the characteristic information of the request signal.
  • the prefetch manager is used to execute the prefetch strategy according to the predicted QoE gain. Select the video block that produces the greatest overall benefit for prefetching to achieve the best optimization effect. For example, the prefetch manager obtains multiple video data associated with the video data to be requested according to the feature information of the request signal, that is, the prefetch manager determines the comprehensive feature by analyzing the feature information of the multiple request signals, and then determines the multiple A video data associated with the video data to be requested is equivalent to that the prefetch manager will determine the video that may be requested by the client in the future according to the request habit of the client.
  • the prefetch manager predicts the maximum benefit value of the associated video data that can be prefetched in the current prefetch period.
  • the prefetch manager needs to predict the maximum benefit value of the associated video data that can be prefetched in the current prefetch period. What is actually predicted here is the benefit that can be maximized.
  • this application gives a specific implementation process as follows:
  • E ⁇ e ⁇ represents the collection of edge servers
  • U ⁇ u ⁇ represents the collection of clients
  • O ⁇ o ⁇ represents the collection of origin servers
  • L E ⁇ l u,e
  • L O ⁇ l e,o
  • l u The transmission capacity of ,e and l e,o can be defined as c u,e (t) and c e,o (t) respectively.
  • the client needs to provide the edge server with the following state information, that is, at time t, the following state information needs to be obtained, including the client's user experience evaluation QoE u (t), and the client's local cache b u (t ), buffer situation ru (t) and end-to-end round trip time
  • the above four indicators can replace QoE u (k), b u (k), ru (k) with video blocks as the granularity.
  • Each video f can be encoded into multiple bitrate versions, which can be expressed as M ⁇ f ⁇ . And each video can be cut into several video blocks, which can be expressed as K ⁇ f ⁇ . Therefore, each video block can have a unique mark, which is expressed as f m,k ,m ⁇ M(f),k ⁇ K(f), using f m,k to represent user experience evaluation.
  • the edge server will use the URL information contained in the request to guess the content that may need to be prefetched, and extract the corresponding state information from the request for subsequent prediction and calculation.
  • the benefit of prefetching a certain video block f m,k is defined as Utility(f m,k ), where the benefit has nothing to do with the corresponding user but only related to the characteristics of the video block itself.
  • the edge server determines the benefit value of the video, where the benefit value is used to evaluate the benefits of prefetching and caching different videos.
  • the probability of rate switching also needs to be considered.
  • the probability of rate switching is related to the length of the video block chain.
  • the DASH video request is composed of a series of continuous video blocks, and when the bit rate of the video block is not switched, the said video chain will be formed. Assuming that the chain length is x, under different chain lengths, the possibility of bit rate switching of the video chain is also different. It can be expressed as p(x). According to the statistics of the chain length of the actual video request, a fitted curve can be obtained as shown in FIG. 8a, which is used to characterize the possibility that the bit rate of the user's video request will not change.
  • the size of the k-th video is defined as c(f m , k+1).
  • a sum of prefetch benefits can be calculated by the following formula, which is as follows:
  • the prefetch manager calculates the maximum benefit value of the associated video data that can be prefetched in a period. It is assumed that it is currently in the first stage of the dth prefetch period and will obtain the information between the edge server and the source server.
  • the available bandwidth is defined as c e,o (dT p ).
  • the weighted average method is used to calculate the maximum number of video blocks that can be prefetched in the current cycle, that is, the number of video blocks that can be prefetched in the d-th cycle can be calculated as follows: It can be calculated that in the first stage, the cumulative amount of prefetchable video blocks is It will be appreciated that prefetch manager calculates a first phase only according to r (dT p) of the real prefetch request to maximize benefit Utility max (dT p), instead of executing prefetch requests immediately.
  • the prefetch manager performs prefetch management on the associated video data according to the maximum benefit value of the associated video data, the current video QOE gain value, and each associated video data.
  • step S603 includes:
  • the prefetch manager determines the video data to be prefetched according to the maximum benefit value of the associated video data and the current video QOE gain value.
  • the prefetch manager first determines the video data to be prefetched according to the maximum benefit value of the associated video data and the current video QOE gain value, that is, only the video with the larger benefit value needs to be determined among all the associated video data. Is the video data to be prefetched.
  • the prefetch manager sends a prefetch request carrying video data to be prefetched to the source server.
  • the prefetch manager Based on the video data to be prefetched determined in step S701, the prefetch manager sends a prefetch request carrying the video data to be prefetched to the source server.
  • the prefetch manager when the prefetch manager sends a prefetch request carrying video data to be prefetched to the source server, it can send the prefetch request based on whether the maximum benefit value of the associated video data exceeds a preset threshold, for example, prefetch
  • the fetch manager predicts that in the first half of the current prefetch cycle, it counts the maximum benefit of prefetch requests in this stage, and does not send prefetch requests to the server. Then, in the second half of the prefetch cycle, the The benefit value of the fetch request is compared with the maximum value obtained in the first half stage. If the maximum value is exceeded, the prefetch request is sent, otherwise, the prefetch request is not sent.
  • the prefetch manager sends the prefetch request on the premise that the above-determined maximum benefit value Utility max (dT p ) is greater than the video benefit value Utility (f m, k ) determined by the edge server.
  • the prefetch manager will execute the benefit-based prefetch strategy, which can be executed according to the following algorithm program: the following algorithm program lines 10 to 13, if a request ends in the first stage After arriving, the prefetch manager will compare the benefit Utility(f m,k ) of this request with the maximum benefit Utility max (dT p ) calculated in the first stage, if it is greater than the Utility(f m,k ) , The request will be executed, that is, request the source server for the content you want to prefetch.
  • Output xf-0 or 1, indicating whether to perform prefetching.
  • the prefetch manager receives the video data to be prefetched returned from the source server, and stores the video data to be prefetched.
  • the prefetch manager stores the video data to be prefetched back from the source server, that is, stores the prefetched video in the local database so that it can be sent to the client immediately upon request. It should be noted that if the video content prefetched by the prefetch manager exceeds the preset duration and is not requested by the client, the prefetch manager can no longer store the prefetched video to ensure that the prefetched video is the latest The hottest video content.
  • the video transmission method provided in this embodiment incorporates the evaluation of the user experience quality gain value into the cache prefetch strategy, improves the rigid strategy of the traditional cache prefetch, and makes it more flexible and effective. It uses the probability of a constant bit rate to calculate and store the corresponding video block. Benefit to make it more in line with the request mechanism of adaptive bit rate video transmission.
  • the behavior mode of users watching videos is traceable.
  • the prefetch strategy set in this application can capture the historical behaviors of users watching videos, predict the video content that the users will request, and then effectively cache and prefetch. With the help of artificial neural network to train the prediction model of QoE gain, while improving the user experience, it effectively improves the reuse of video content.
  • the following describes an embodiment of the video transmission method on the side where the execution subject is the client.
  • an embodiment of the present application provides a video transmission method, which involves a specific process of a client requesting video data. As shown in FIG. 10, the method includes:
  • the client sends a request signal to the source server; the request signal is used to instruct to obtain video data to be requested.
  • the client sends a request information number to the source server, where the request signal is used to instruct to obtain the video data to be requested.
  • the request signal may be received by the client through an external input device or a touch screen input device.
  • the signal generated after the input instruction does not limit the specific form of the input device in this embodiment.
  • the client receives the video data to be requested; the data to be requested is sent back by the edge server according to the request signal redirected by the source server.
  • the origin server will redirect the request signal to the edge server, and the edge server will return the video data to be requested by the client to the client.
  • the client receives The data to be requested is returned by the edge server according to the request signal redirected by the origin server.
  • this application adds an edge server to the entire system to use the computing power and storage space at the edge of the network to accelerate video transmission, thereby optimizing the user's viewing experience.
  • the internal structure diagram of the source server, the client, and the edge server may be the internal structure diagram of the computer device as shown in FIG. 11.
  • the computer equipment includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus.
  • the processor of the computer device is used to provide calculation and control capabilities.
  • the memory of the computer device includes a non-volatile storage medium and an internal memory.
  • the non-volatile storage medium stores an operating system and a computer program.
  • the internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium.
  • the network interface of the computer device is used to communicate with an external terminal through a network connection.
  • the computer program is executed by the processor to realize a video transmission method.
  • the display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen
  • the input device of the computer equipment can be a touch layer covered on the display screen, or it can be a button, a trackball or a touchpad set on the housing of the computer equipment , It can also be an external keyboard, touchpad, or mouse.
  • FIG. 11 is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied.
  • the specific computer device may Including more or fewer parts than shown in the figure, or combining some parts, or having a different arrangement of parts.
  • a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:
  • the edge server receives the request signal sent by the client, and extracts the characteristic information of the request signal; the request signal is used to instruct to obtain the video data to be requested;
  • the edge server searches the local database for the video data to be requested according to the characteristic information of the request signal; the local database stores prefetched videos and cached videos;
  • the edge server transmits the video data to be requested to the client.
  • the client sends a request signal to the source server; the request signal is used to instruct to obtain the video data to be requested;
  • the client receives the video data to be requested; the data to be requested is returned by the edge server according to the request signal redirected by the source server.
  • Non-volatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
  • Volatile memory may include random access memory (RAM) or external cache memory.
  • RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Channel (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

Abstract

La présente invention concerne un procédé et un système de transmission vidéo, et un support de données. Le procédé comprend les étapes suivantes : un serveur périphérique reçoit un signal de demande envoyé par un client et extrait des informations de caractéristique du signal de demande, le signal de demande étant utilisé pour indiquer d'obtenir des données vidéo à demander ; le serveur périphérique recherche, selon les informations de caractéristique du signal de demande, une base de données locale pour les données vidéo à demander, la base de données locale stocke des vidéos pré-extraites et des vidéos mises en cache ; le serveur périphérique transmet les données vidéo à demander au client.
PCT/CN2020/096239 2019-06-19 2020-06-16 Procédé et système de transmission vidéo, et support de stockage WO2020253664A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910530061.1 2019-06-19
CN201910530061.1A CN110430440B (zh) 2019-06-19 2019-06-19 视频传输方法、系统、计算机设备和存储介质

Publications (1)

Publication Number Publication Date
WO2020253664A1 true WO2020253664A1 (fr) 2020-12-24

Family

ID=68408672

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/096239 WO2020253664A1 (fr) 2019-06-19 2020-06-16 Procédé et système de transmission vidéo, et support de stockage

Country Status (2)

Country Link
CN (1) CN110430440B (fr)
WO (1) WO2020253664A1 (fr)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115988267A (zh) * 2022-12-20 2023-04-18 哈尔滨工业大学（深圳） 一种基于dash的视频码率自适应调整方法及系统
CN116112740A (zh) * 2023-01-19 2023-05-12 深圳大学 一种视频播放方法、装置、电子设备及存储介质

Families Citing this family (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110430440B (zh) * 2019-06-19 2021-11-30 鹏城实验室 视频传输方法、系统、计算机设备和存储介质
CN111212301B (zh) * 2020-01-09 2021-10-08 鹏城实验室 一种视频码率的匹配方法、存储介质及终端设备
CN111372096B (zh) * 2020-03-12 2022-02-18 重庆邮电大学 一种基于d2d辅助的视频质量自适应缓存方法和设备
CN111966889B (zh) * 2020-05-20 2023-04-28 清华大学深圳国际研究生院 一种图嵌入向量的生成方法以及推荐网络模型的生成方法
CN112202802B (zh) * 2020-10-10 2021-10-01 中国科学技术大学 C-ran架构中基于强化学习的vr视频多级缓存方法和系统
CN112202800B (zh) * 2020-10-10 2021-10-01 中国科学技术大学 C-ran架构中基于强化学习的vr视频边缘预取方法和系统
CN112565606B (zh) * 2020-12-02 2022-04-01 鹏城实验室 全景视频智能传输方法、设备及计算机存储介质
CN113301396B (zh) * 2021-05-12 2023-01-31 中国联合网络通信集团有限公司 基于边缘计算的视频服务方法及边缘计算服务器
CN113596021B (zh) * 2021-07-28 2023-02-07 中国人民解放军国防科技大学 一种支持神经网络的流媒体码率自适应方法、装置和设备
CN114143376A (zh) * 2021-11-18 2022-03-04 青岛聚看云科技有限公司 一种用于加载缓存的服务器、显示设备及资源播放方法
CN114584801B (zh) * 2022-01-13 2022-12-09 北京理工大学 一种基于图神经网络推荐算法的视频资源缓存方法

Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103916698A (zh) * 2014-04-24 2014-07-09 梁科 视频传输方法、设备及系统
CN104796443A (zh) * 2014-01-17 2015-07-22 中兴通讯股份有限公司 一种移动流媒体用户体验质量QoE修正方法和服务器
WO2016109916A1 (fr) * 2015-01-05 2016-07-14 华为技术有限公司 Appareil de prédiction de qualité d'expérience (qoe), dispositif de réseau et procédé
CN105812834A (zh) * 2016-05-10 2016-07-27 南京大学 基于聚类信息的视频推荐服务器、推荐方法和预缓存方法
CN106027291A (zh) * 2015-11-25 2016-10-12 北京邮电大学 基于韦伯费希纳定理的BE业务QoE评价方法
US9519614B2 (en) * 2012-01-10 2016-12-13 Verizon Digital Media Services Inc. Multi-layer multi-hit caching for long tail content
CN106303704A (zh) * 2016-08-19 2017-01-04 上海交通大学 一种基于代理服务器的dash流媒体直播系统及方法
CN109218814A (zh) * 2018-09-28 2019-01-15 西安交通大学 一种云计算环境下QoE驱动的HAS直播频道调度方法
CN109327867A (zh) * 2018-10-26 2019-02-12 西安交通大学 LTE网络下QoE驱动的视频码率自适应和资源分配联合算法
CN110430440A (zh) * 2019-06-19 2019-11-08 鹏城实验室 视频传输方法、系统、计算机设备和存储介质

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108235131B (zh) * 2018-01-30 2020-07-03 重庆邮电大学 一种基于dash的全景视频自适应传输方法

Patent Citations (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9519614B2 (en) * 2012-01-10 2016-12-13 Verizon Digital Media Services Inc. Multi-layer multi-hit caching for long tail content
CN104796443A (zh) * 2014-01-17 2015-07-22 中兴通讯股份有限公司 一种移动流媒体用户体验质量QoE修正方法和服务器
CN103916698A (zh) * 2014-04-24 2014-07-09 梁科 视频传输方法、设备及系统
WO2016109916A1 (fr) * 2015-01-05 2016-07-14 华为技术有限公司 Appareil de prédiction de qualité d'expérience (qoe), dispositif de réseau et procédé
CN106027291A (zh) * 2015-11-25 2016-10-12 北京邮电大学 基于韦伯费希纳定理的BE业务QoE评价方法
CN105812834A (zh) * 2016-05-10 2016-07-27 南京大学 基于聚类信息的视频推荐服务器、推荐方法和预缓存方法
CN106303704A (zh) * 2016-08-19 2017-01-04 上海交通大学 一种基于代理服务器的dash流媒体直播系统及方法
CN109218814A (zh) * 2018-09-28 2019-01-15 西安交通大学 一种云计算环境下QoE驱动的HAS直播频道调度方法
CN109327867A (zh) * 2018-10-26 2019-02-12 西安交通大学 LTE网络下QoE驱动的视频码率自适应和资源分配联合算法
CN110430440A (zh) * 2019-06-19 2019-11-08 鹏城实验室 视频传输方法、系统、计算机设备和存储介质

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115988267A (zh) * 2022-12-20 2023-04-18 哈尔滨工业大学（深圳） 一种基于dash的视频码率自适应调整方法及系统
CN115988267B (zh) * 2022-12-20 2023-09-15 哈尔滨工业大学（深圳） 一种基于dash的视频码率自适应调整方法及系统
CN116112740A (zh) * 2023-01-19 2023-05-12 深圳大学 一种视频播放方法、装置、电子设备及存储介质

Also Published As

Publication number Publication date
CN110430440B (zh) 2021-11-30
CN110430440A (zh) 2019-11-08

Similar Documents

Publication Publication Date Title
WO2020253664A1 (fr) Procédé et système de transmission vidéo, et support de stockage
US20160142510A1 (en) Cache-aware content-based rate adaptation mechanism for adaptive video streaming
CN103108257B (zh) 一种用于嵌入式终端改善流媒体播放质量的方法及系统
US20140089467A1 (en) Content stream delivery using pre-loaded segments
CN108366089B (zh) 一种基于内容流行度和节点重要度的ccn缓存方法
Jiang et al. Q-FDBA: improving QoE fairness for video streaming
CN111491331B (zh) 一种雾计算网络中基于迁移学习的网络感知自适应缓存方法
Narayanan et al. Making content caching policies' smart'using the deepcache framework
Wang et al. Deepchunk: Deep q-learning for chunk-based caching in wireless data processing networks
CN112752308B (zh) 一种基于深度强化学习的移动预测无线边缘缓存方法
Zhao et al. Popularity-based and version-aware caching scheme at edge servers for multi-version VoD systems
CN109951317B (zh) 一种基于用户驱动的流行度感知模型的缓存替换方法
CN114786034A (zh) 一种基于边缘计算的全景视频优化传输方法和系统
US20220191260A1 (en) Method for playing on a player of a client device a content streamed in a network
Darwich et al. Edge computing for efficient storage and low-latency video streaming in cloud environments
Shi et al. CoLEAP: Cooperative learning-based edge scheme with caching and prefetching for DASH video delivery
CN109195180A (zh) 一种减小移动内容中心网络中内容获取时延的解决方法
Fan et al. Pa-cache: Learning-based popularity-aware content caching in edge networks
Osman et al. A comparison of cache replacement algorithms for video services
CN113766540B (zh) 低时延的网络内容传输方法、装置、电子设备及介质
Ramu et al. A study on web prefetching techniques
Aguilar-Armijo et al. Segment prefetching at the edge for adaptive video streaming
Xu et al. An incremental learning based edge caching system: From modeling to evaluation
Duraimurugan et al. Maximizing the quality of service in distributed multimedia streaming in heterogeneous wireless network
Meint et al. From FIFO to predictive cache replacement

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20827054

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20827054

Country of ref document: EP

Kind code of ref document: A1